Author's personal copy Journal of Chromatography A, 1216 (2009) 1140–1146

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Aqueous normal-phase retention of nucleotides on silica hydride columns Joseph J. Pesek a,∗ , Maria T. Matyska a , Milton T.W. Hearn b , Reinhard I. Boysen b a b

Department of Chemistry, San Jose State University, San Jose, CA 95112, USA Australian Research Council Special Research, Centre for Green Chemistry, Monash University, Clayton, Victoria 3800, Australia

a r t i c l e

i n f o

Article history: Received 16 October 2008 Received in revised form 15 December 2008 Accepted 17 December 2008 Available online 25 December 2008 Keywords: HPLC Polar compound retention Nucleotides

a b s t r a c t The use of silica hydride-based stationary phases for the retention and analysis of nucleotides has been investigated. Both reversed-phase columns with a hydride surface underneath as well as those with an unmodified or a minimally modified hydride material were tested. With these systems, an aqueous normal-phase mode was used with high organic content mobile phases in combination with an additive to control pH for the retention of the hydrophilic nucleotides. Isocratic and gradient elution formats have been used to optimize separations for mixtures containing up to seven components. All conditions developed are suitable for methods that utilize mass spectrometry detection. © 2008 Elsevier B.V. All rights reserved.

1. Introduction Nucleotides are important phosphate containing compounds that are found in living cells and are associated with a vast array of metabolic and biological processes. They play key roles in the synthesis of DNA and RNA, are involved in signal transduction pathways, function as coenzymes in biosynthetic processes and serve as energy reservoirs in biological systems [1–6]. Thus improved methods of analysis for these compounds are of continued interest, especially techniques which can distinguish among the nucleotides based on their degree of phosphorylation. Typically nucleotides are often separated by ion-exchange chromatography [7–9] or in some instances by reversed-phased HPLC methods [7,10–14]. However, due to the high polarity of most nucleotides retention is generally low with most C18 stationary phases. Thus, ion-exchange methods are more amenable to the hydrophilic nature of these compounds. Samples of biological importance isolated from complex matrices are now frequently analyzed utilizing mass spectrometry as the means of detection. The anion-exchange methods developed for the separation of nucleotides are generally not suitable for MS detection due to the high concentrations of buffers and salts which are usually not volatile. Thus newer approaches which are compatible with either optical or MS detection are desirable. Silica hydride-based HPLC stationary phases were first proposed as an alternative to other types of silica materials more than 10 years ago. A large number of studies have confirmed that they possess

∗ Corresponding author. E-mail address: [email protected] (J.J. Pesek). 0021-9673/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2008.12.046

both high stability and a broad range of chromatographic properties that are unique when compared to other separation media [15–18]. One of the desirable features that has been identified for hydride separation materials is their ability to be used over a wide range of mobile phase compositions from 100% aqueous to pure nonpolar organic solvents. Therefore, these stationary phases can function in high water (reversed-phase), high organic with some water present (aqueous normal-phase) and pure organic (organic normal-phase) mobile phases. The reversed-phase (RP) and aqueous normal-phase (ANP) modes are highly complementary because water is the common component in the mobile phase. With silica hydride-based stationary phases it is possible to rapidly change from RP to ANP due to the rapid equilibration of these separation materials and in some cases both mechanisms operate simultaneously thus retaining both hydrophobic (by reversed-phase) and hydrophilic (by aqueous normal phase) compounds in a single isocratic run [17]. There have been a number of silica hydride-based stationary phases studied so it is now known that the relative amount of reversed-phase and aqueous normal-phase capabilities can be adjusted by varying the surface composition [17]. The surface composition that determines the retention properties is a combination of the base silica hydride and the organic moiety attached to it. This investigation has utilized several types of silica hydride separation materials in order to determine the type and extent of organic moiety surface coverage that would provide the best ANP retention capabilities for nucleotides. The designation of ANP for the retention of polar compounds, such as nucleotides, is used in this report to distinguish silica hydride materials from HILIC stationary phases. While silica hydride phases can make the transition from ANP to RP by the addition of water to the mobile phase HILIC materials do not have this capability and thus can only retain polar compounds. In addition,

Author's personal copy J.J. Pesek et al. / J. Chromatogr. A 1216 (2009) 1140–1146

the experimental conditions in the methods reported are suitable for detection by either optical or mass spectroscopic techniques. 2. Experimental 2.1. Materials The undecenoic acid (UDA) stationary phase was synthesized in house and packed into a 150 mm × 4.6 mm column. The other silica hydride stationary phases used in this study were the Cogent Type-C columns in 150 mm × 1.0 mm, 100 mm × 2.1 mm and 75 mm × 4.6 mm I.D. from MicroSolv Technology (Eatontown, NJ, USA). The Diamond Hydride (DH) phase contains a small amount of an organic moiety (∼2% carbon as reported by the manufacturer) on a silica hydride surface. All hydride columns were made with 4.0 ␮m silica (Astrosil, Stellar Phases, Langehorn, PA, USA). A common commercial C18 column (Agilent Eclipse, Wilmington, DE, USA, 4.6 mm × 150 mm I.D.) was included in initial screening tests to provide a comparison for the reversed-phase hydride materials. The nucleotides adenosine-3 ,5 -cyclic monophosphate; adenosine-5 -monophosphate; adenosine-5 -triphosphate; thymidine-5 -triphosphate; uridine-5 -triphosphate; cytosine-5 triphosphate; guanosine-5 -triphosphate were purchased from Sigma–Aldrich (Milwaukee, WI, USA). Samples for UV detection were made at 100 ␮g/mL and for MS detection at 10 ␮g/mL. Solvents and buffer components for the mobile phase were obtained in the highest purity available. 2.2. Instrumentation All HPLC investigations utilized an Agilent (Little Falls, DE, USA) 1100 Series LC system, including degasser, binary pump, temperature-controlled autosampler and temperature-controlled column compartment. The UV detector was a diode array (DAD) system (Agilent Model DAD SL G1315C). For LC–MS, the mass spectrometer was an Agilent (Santa Clara, CA, USA) Model 6210 MSD time-of-flight (TOF) with a dual sprayer electrospray source (ESI). 2.3. Methods The undecenoic stationary phase was synthesized using a process that has been previously reported [19,20]. The column was packed into a 75 mm × 4.6 mm I.D. column using a 90:10 carbon tetrachloride/methanol slurry with methanol as the driving solvent. Stock solutions of the nucleotides were made in deionized (DI) water in the range of 0.2–0.7 mg/mL. Sample solutions were made by diluting the stock in 50:50 acetonitrile/water containing 0.1% ammonium formate. The mobile phase organic solvent was composed of 0.1% ammonium formate in acetonitrile. Water containing 0.1% ammonium formate made up the difference. The column flow rate was 1.0 mL/min. The column temperature was 20 ◦ C. The gradients used in this study are designated in each figure caption. For gradient experiments the flow rate was 1.0 mL/min with UV detection and 0.4 mL/min for MS operation. The equilibration time between successive runs was 5 min. For repeatability studies, the number of replicates performed under each condition tested was noted. 3. Results and discussion 3.1. Reversed-phase sorbents under ANP conditions It has been demonstrated that a hydride-based column designed for reversed-phase through the bonding of a hydrophobic moiety such as C18 can also function in the ANP mode

1141

[17]. Most columns based on ordinary silica do not have this capability. In order to test this concept, a mixture of seven nucleotides (adenosine-3 ,5 -cyclic monophosphate; adenosine-5 monophosphate; adenosine-5 -triphosphate; thymidine-5   triphosphate; uridine-5 -triphosphate; cytosine-5 -triphosphate; guanosine-5 -triphosphate) was analyzed on several columns with typical reversed-phase properties. The first examined was a commercial C18 based on organosilane bonding of octadecyl groups to silica. Fig. 1A shows the result of testing this column using a type of ANP gradient that was successful in the retention and separation of other polar compounds [21]. As seen in the figure, all of the components of this mixture co-elute at essentially the void volume of the column. Using the same gradient on a shorter hydride-based C18 column produced some retention of the components as seen (Fig. 1B) as a separation between the baseline fluctuations at the void volume and the two peaks for the nucleotides. While some retention was achieved on the hydride-based C18, the selectivity was only partially sufficient to resolve some of the components in the mixture under these mobile phase conditions. A third column, a hydride-based cholesterol-bonded phase, with significant RP properties was also tested using a slightly longer gradient profile. Under these conditions more retention of the nucleotides (Fig. 1C) and somewhat better resolution of the four components were achieved. The longer gradient on the standard silica-based C18 column showed neither improved retention nor any resolution of the components while the hydride-based C18 had longer retention but no significant improvement of resolution. For nucleotides, the hydride-based stationary phases with hydrophobic groups produce retention under ANP conditions, but the selectivity was not high, at least under the conditions tested in this study. 3.2. Diamond Hydride column The Diamond Hydride stationary phase is essentially a silica hydride surface with a low coverage of organic moiety attached. A previous study has demonstrated that this sorbent can be used for the separation of metabolites such as amino acids, carbohydrates and small organic acids in the ANP mode [21]. Thus, it is a good choice for evaluation with nucleotides. A number of monophosphate and triphosphate nucleotides were analyzed under isocractic conditions using acetonitrile/water mobile phases at high organic content to achieve ANP conditions. In addition, various mobile phase additives were evaluated including acetic acid, formic acid, ammonium formate and ammonium acetate. Ammonium formate at 0.1% (w/v) consistently gave the best peak shape but in order to reduce the analysis time, and to produce better peak symmetry and efficiency it was necessary to employ gradient elution conditions. Fig. 2 shows the separation of a mixture of two monophosphate and two triphosphate nucleotides using a simple linear gradient from 95% to 70% acetonitrile over 10 min. Under these conditions the total analysis time was less than 10 min and the peak shape is good. Two small impurities, possibly other triphosphate nucleotides, are also seen in the chromatogram. These peaks could be due to either contamination or degradation but no attempt was made to identify them. As shown in Fig. 2, the most strongly retained compounds were the triphoshates (as predicted by the ANP mechanism) and some tailing can occur depending on the mobile phase additive and gradient selected. Fig. 3A shows an example of the separation of the three adenosine analytes using a gradient that was only slightly different to the one employed to achieve the results shown in Fig. 2. The gradient starts at 90% acetonitrile in the mobile phase but also goes to 70% over 10 min thus making it not as steep as the gradient used for the chromatogram in Fig. 2. The last component, adenosine 5 triphosphate, displays noticeable tailing in comparison to the first two compounds. An improvement in peak shape can be achieved

Author's personal copy 1142

J.J. Pesek et al. / J. Chromatogr. A 1216 (2009) 1140–1146

Fig. 2. Separation of nucleotide mixture on Cogent Diamond Hydride column. Dimensions: 2.1 mm × 100 mm. Flow rate = 0.3 mL/min Detection at 254 nm. Gradient: 0.0 min 95% B, 0.0–10.0 min to 70% B. Mobile phase conditions are the same as in Fig. 1. Solutes: 1 = adenosine-3 ,5 -cyclic monophosphate; 2 = adenosine-5 monophosphate; 3 = uridine-5 -triphosphate; 4 = cytosine-5 -triphosphate.

by adding a small amount of ammonia to the sample (Fig. 3B). The amount used here (5 ␮L of 12% ammonia/mL) does not appreciably affect the retention times. The improvement of peak shape by the addition of ammonia is consistent with previous results for various organic acids and amino acids where efficiency and peak symmetry are dependent upon the acidity of the solution [21]. In particular, peak shape improved considerably for organic acids such as fumaric, maleic and citric as the pH was raised by changing the mobile phase from formic acid to ammonium formate and by including NH4 OOCH in the sample solvent. Under more acidic conditions it is likely that a mixed retention mechanism takes places as both protonated and unprotonated acid solute species are present. These results suggest that mass transfer of the solute between the mobile phase and the stationary phase is improved as the solution conditions favor the greatest amount of unprotonated species. 3.3. Undecenoic acid column The retention capabilities of the UDA column are the result of both hydrophilic (hydride surface and carboxylic acid functionality) and hydrophobic (alkyl chain of bonded acid) properties. With its greater hydrophobic nature, retention and selectivity for nucleotides on the UDA column should be different than the Diamond Hydride in the ANP mode. These differences are illustrated in Fig. 4 for the separation of seven nucleotides on the UDA column under isocratic conditions. At 80% ACN in the mobile phase (Fig. 4A), good retention was obtained with one pair of analytes (thymidine triphosphate and adenosine triphosphate) not resolved. If the organic content of the mobile phase was raised to 85%, then retention became longer and the two previously unresolved components were separated. This result is in contrast to the data obtained with the Diamond Hydride column where complete resolution of this mixture was not successful under isocratic conditions and required Fig. 1. Chromatograms for gradient elution of a seven-component nucleotide mixture on reversed-phased columns. (A), Agilent Eclipse, (B), Cogent Bidentate C18 and (C) Cogent UDC Cholesterol. Detection at 254 nm. Gradient for A and B: 0.0 min 90% B, 0.0–10.0 min to 70% B. Gradient for C: 0.0–0.5 min 95% B, 0.5–20 min to 30% B, 20.0–25.0 min hold 30% B. Mobile phase: A, DI water + 0.1% ammonium

formate; B, 90% acetonitrile/10% DI water + 0.1% ammonium formate. Solutes: 1 = adenosine-3 ,5 -cyclic monophosphate; 2 = adenosine-5 -monophosphate; 3 = adenosine-5 -triphosphate; 4 = thymidine-5 -triphosphate; 5 = uridine-5 triphosphate; 6 = cytosine-5 -triphosphate; 7 = guanosine-5 -triphosphate.

Author's personal copy J.J. Pesek et al. / J. Chromatogr. A 1216 (2009) 1140–1146

Fig. 3. Separation of three adenosine compounds on the Cogent Diamond Hydride column. (A), Sample without ammonia and (B), sample with ammonia. Column and conditions same as Fig. 2. Gradient conditions are the same as in Fig. 1A. Solutes: 1 = adenosine-3 ,5 -cyclic monophosphate; 2 = adenosine-5 monophosphate; 3 = adenosine-5 -triphosphate.

1143

Fig. 4. Isocratic separation of seven-component mixture of nucleotides on the undecenoic acid column. (A), 80:20 90% acetonitrile/10% DI water + 0.1% ammonium formate/DI water + 0.1% ammonium formate and (B), 85:15 90% acetonitrile/10% DI water + 0.1% ammonium formate/DI water + 0.1% ammonium formate. Column: 4.6 mm × 75 mm. Flow rate 1 mL/min. Detection at 254 nm. Solutes same as Fig. 1.

Author's personal copy 1144

J.J. Pesek et al. / J. Chromatogr. A 1216 (2009) 1140–1146

Fig. 5. Gradient separation of seven-component mixture of nucleotides on the undecenoic acid column. Column and solutes are the same as given in the legend to Fig. 4, gradient same as Fig. 1A.

a gradient. The improved resolution is most likely due to the enhanced interaction of the nitrogen sites on the nucleotides with the stationary phase provided by the carboxylic acid functionality. The analysis time with the UDA column can be reduced but with some loss in resolution by using the same gradient as in Fig. 1A. The small shoulder on peak 5 is likely the small peak (identity not determined) in Fig. 4. The results under these experimental conditions are shown in Fig. 5. While this may not be a practical solution when using UV detection as shown, if quantitative determinations are necessary, such resolution is certainly practical for MS detection. Adequate retention of all the compounds was obtained and the m/z values for the (M−H)− ions of each of these compounds, generated by ESI under these mobile phase conditions, were sufficiently different so that unambiguous identification can be made via the extracted ion chromatogram (EIC). The closest values in m/z for the compounds in this mixture differ by 2 amu (480 and 482) which is more than sufficient for even a single quadrupole MS to resolve at this mass/charge ratio. The presence of ammonia in the sample solution also has a positive effect on peak shape for the UDA column in a manner similar to that observed for the DH column described above (Fig. 3B). Fig. 6 shows a comparison of the separation of five nucleotides using the same gradient where the only difference was that the sample solution was without (Fig. 6A) or with (Fig. 6B) ammonia added. A distinct improvement in peak shape and resolution was obtained for the three triphosphates which were poorly resolved when there was no ammonia in the sample solution. In this case retention increased slightly due to the fact that a higher amount (twice the amount used in Fig. 3B) shifts the sample further to the more unprotonated state and may also have resulted in greater ionization of the carboxylic acid functionality of the bonded moiety. The process of ANP retention on the hydride-based phases is not fully understood but more detailed investigations are currently underway to elucidate the exact mechanism. A comparison was also made between the UDA and DH columns using flow rate conditions that gave approximately equivalent retention times. All other conditions, gradient, mobile phase components and sample solvent, were held the same for both columns. This comparison is shown in Fig. 7. As can be seen resolution and efficiency are better on the UDA column (Fig. 7B) than on the DH column (Fig. 7A). This result does not imply that the UDA is superior under all conditions or that the DH separation of this mixture could not be improved by further method optimization.

Fig. 6. Gradient separation of five-component mixture of nucleotides on the undecenoic acid column. (A), Sample without ammonia and (B), sample with ammonia. Gradient same as Fig. 1A. Other conditions are the same as given in the legend to Fig. 4. Solutes: 1 = adenosine-3 ,5 -cyclic monophosphate; 2 = adenosine5 -monophosphate; 3 = adenosine-5 -triphosphate; 4 = thymidine-5 -triphosphate; 5 = uridine-5 -triphosphate.

A final test on the UDA column involved the repeatability of consecutive sample injections which is shown in Fig. 8. Five representative runs of the seven-component nucleotide sample run under isocratic conditions on the UDA column are shown. As can be seen, the chromatograms are highly repeatable and thus the column provided stable operation for consecutive analyses under the same experimental conditions. If a gradient method was used, consecutive runs also resulted in chromatograms with retention times for the same peak with RSD values